Premium
[O4–04–02]: MULTIMODAL IMAGING APPROACH TO UNDERSTAND THE IMPACT OF VASCULAR HEALTH ON BRAIN DIFFUSION, PERFUSION, METABOLISM, AND STRUCTURE
Author(s) -
Vemuri Prashanthi,
Knopman David S.,
Lowe Val,
Przybelski Scott A.,
GraffRadford Jonathan,
Senjem Matthew L.,
Reid Robert I.,
Schwarz Christopher G.,
Gunter Jeffrey L.,
Machulda Mary M.,
Petersen Ronald C.,
Jack Clifford R.
Publication year - 2017
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1016/j.jalz.2017.07.437
Subject(s) - fractional anisotropy , medicine , cardiology , cerebral blood flow , perfusion , perfusion scanning , diffusion mri , diabetes mellitus , stroke (engine) , coronary artery disease , pittsburgh compound b , pathology , dementia , disease , endocrinology , magnetic resonance imaging , radiology , mechanical engineering , engineering
Background:There is increasing evidence that cerebrovascular disease (CVD) plays a role in Alzheimer’s disease (AD) pathogenesis. Neuroimaging studies typically measure CVD markers and neurodegeneration and then treat them as separate factors. Clinical manifestations of AD likely result from a combination of CVD and neurodegeneration, which varies across individuals. We created a single MRI-based phenotype score that captures the relative contribution AD-related CVD and neurodegeneration.Methods:In a community-based study of aging and dementia, MRI scans were acquired on 1,333 participants (mean age1⁄477.3+6.4,62% women), in two independent samples (n1⁄4768 and n1⁄4564) and analyzed for two cerebrovascular markers (white matter hyperintensity[WMH] volume, presence of infarct) and two neurodegeneration markers (hippocampal volume, cortical thickness). We used linear regression to examine the relative independent relationship of each with age-adjusted memory scores in the first sample. We computed the phenotype score by taking the linear combination of the four MRI markers weighted by their relative contribution to memory, and tested the extent to which this score related to memory in the second sample and to diagnosis in a combined analysis. We further validated the phenotype among 42 participants who came to autopsy. Results:Based on b weights from the linear regression, the phenotype was derived with the following equation: CVD-AD phenotype1⁄4-0.067*WMH volume-0.031*infarct presence+0.00018*hippocampus volume+0.76*cortical thickness. This CVD-AD score was strongly related to memory in the first, second, and combined samples (all ps<0.001). Those with AD(n1⁄442) had lower scores than those with MCI(n1⁄4200), who had lower scores than controls (n1⁄4766;F1⁄446.78,p<0.001). The phenotype was strongly related to pathology measures of neurofibrillary tangles(r1⁄4-0.435,p1⁄40.004), neuropil threads(r1⁄4-0.381,p1⁄40.001), neuronal loss(r1⁄40.361,p1⁄40.023), atrophy(r1⁄4-0.367,p1⁄40.017), and small vessel cerebrovascular disease(r1⁄40.308,p1⁄40.042). Conclusions:We derived a single phenotype that reflects the combination of AD-related CVD and neurodegeneration weighted by their relative importance to memory. We validated the score in an independent sample, and by showing its relationship with diagnostic category and neuropathology. In addition to its utility as an outcome or predictor, our findings suggest that both CVD and neurodegeneration are core features of AD and that the expression of the disease clinically in an individual may be due to the unique combination of the two features.